TRAINING NEEDS ANALYSIS ON AI UTILIZATION TO OPTIMIZE DEEP LEARNING FOR ELEMENTARY TEACHERS IN SUMEDANG

Abstract

The introduction of the deep learning approach through Permendikdasmen Number 13 of 2025 has resulted in ever-increasing expectations being placed on elementary school teachers in Indonesia to integrate artificial intelligence (AI) into classroom practice, but training needs that would facilitate this have not been well documented at the district level. This kind of research which is conducted is focusing on the training needs of elementary school teachers in Sumedang Regency, West Java in using AI to optimize Deep Learning approach including the gaps and needs related to competence as well as training and assistant, Barriers that existed no obstacle Private Training Methods priority content. Methods: We performed a cross-sectional quantitative survey using an online questionnaire containing five-point Likert-scale items and open-ended questions with 233 elementary school teachers in Sumedang Regency, which were then analyzed through descriptive statistics and keyword-based content analysis. Results indicated high levels of competency gap: while teachers were moderately familiar with AI (M = 3.70) and Deep Learning (M = 3.42), almost all teachers rated themselves as highly in need of technical support (M = 4.11) and training (M = 4.24). Limited knowledge/competence (30.5%) and inadequate infrastructure (23.2%) were the most frequent barriers to training. Teachers had a stronger preference for an offline, practice-based training (47.2%), rather than fully online (6.9%), and most commonly requested that content would include information about general AI/Deep Learning concepts and ease-of-use practice with continued mentorship over time. Conclusively, these findings offer a foundation for developing an organized and practice-based AI training series for elementary school teachers in the Regency of Sumedang.